

\documentclass[11pt]{article}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\usepackage{natbib}
\usepackage{eurosym}
\usepackage{amsfonts,amsmath}
\usepackage{amssymb}
\usepackage{hyperref}
\usepackage{setspace}
\usepackage{graphicx}
\usepackage{multirow}
\usepackage{pdflscape}
\usepackage{rotating}
%\usepackage{caption}
\usepackage{tikz}
\usepackage[normalem]{ulem}
\usepackage{chngpage}
\usepackage{booktabs}
\usepackage{multirow}
\usepackage[english]{babel}
\usepackage{tabularx}
\usepackage{array}
\usepackage[center]{caption}
\usepackage{pdflscape}
\usepackage{rotating}
\usepackage{url}





\setcounter{MaxMatrixCols}{10}
%TCIDATA{OutputFilter=LATEX.DLL}
%TCIDATA{Version=5.00.0.2552}
%TCIDATA{<META NAME="SaveForMode" CONTENT="1">}
%TCIDATA{LastRevised=Monday, August 31, 2020 10:58:33}
%TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
%TCIDATA{Language=American English}

\newtheorem{theorem}{Theorem}
\newtheorem{acknowledgement}[theorem]{Acknowledgement}
\newtheorem{algorithm}[theorem]{Algorithm}
\newtheorem{axiom}[theorem]{Axiom}
\newtheorem{case}[theorem]{Case}
\newtheorem{claim}[theorem]{Claim}
\newtheorem{conclusion}[theorem]{Conclusion}
\newtheorem{condition}[theorem]{Condition}
\newtheorem{conjecture}[theorem]{Conjecture}
\newtheorem{corollary}[theorem]{Corollary}
\newtheorem{criterion}[theorem]{Criterion}
\newtheorem{definition}[theorem]{Definition}
\newtheorem{example}[theorem]{Example}
\newtheorem{exercise}[theorem]{Exercise}
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{notation}[theorem]{Notation}
\newtheorem{problem}[theorem]{Problem}
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{remark}[theorem]{Remark}
\newtheorem{solution}[theorem]{Solution}
\newtheorem{summary}[theorem]{Summary}
\newenvironment{proof}[1][Proof]{\noindent\textbf{#1.} }{\ \rule{0.5em}{0.5em}}
%\input{tcilatex}

\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}

\begin{document}

\author{Raymond Fisman, Pamela Jakiela, Shachar Kariv and Silvia Vannutelli%
\thanks{%
Fisman: Boston University (rfisman@bu.edu); Jakiela: Center for Global
Development (pjakiela@cgdev.org); Kariv: University of California, Berkeley
(kariv@berkeley.edu); Vannutelli: Northwestern University (silvia.vannutelli@northwestern.edu).}}
\title{Replication Material for: The Distributional Preferences \\
of Americans, 2013-2016\thanks{%
We are grateful to Ernst Fehr and Daniel Markovits for helpful discussions
and for suggestions from the audience at the Ben-Porath Annual Lecture at
the Hebrew University of Jerusalem and a number of seminar participants. We
thank the American Life Panel (ALP) team at the RAND Corporation for
technical and administrative support. We acknowledge financial support from
the Center for Equitable Growth (CEG) at the University of California,
Berkeley. Any findings, opinions and conclusions expressed in this material
are those of the authors and do not necessarily reflect the views of the
institutions or any funding agency.{}}{}}
\date{\today}
\maketitle

\begin{abstract}
We study the distributional preferences of Americans during 2013-2016, a
period of social and economic upheaval. We decompose preferences into two
qualitatively different tradeoffs -- fair-mindedness versus self-interest, and
equality versus efficiency -- and measure both at the individual level in a
large and diverse sample. Although Americans are heterogeneous in terms of
both fair-mindedness and equality-efficiency orientation, we find that the
individual-level preferences in 2013 are highly predictive of those in 2016.
Subjects that experienced an increase in household income became more
self-interested, and those who voted for Democratic presidential candidates
in both 2012 and 2016 became more equality-oriented.

\noindent \textbf{JEL Classification Numbers:} C79, C91, D63.

\noindent \textbf{Keywords}: distributional preferences,
social preferences, fairness, impartiality, equality, efficiency,
rationality, revealed preference, redistribution, political decisions,
voting, household income, American Life Panel (ALP), experiment.
\end{abstract}

\newpage


\newpage \clearpage

\begin{figure}[htbp!]
	\caption{ The relationship between the log-price ratio ($\log
		(p_{s}/p_{o})$) and the token share ($\pi _{s}/(\pi _{s}+\pi _{o})$)for
		selected subjects.}
	\begin{center}
		\includegraphics[width=1\textwidth]{tables-and-figures/figure1.pdf}

	\end{center}
	\label{fig:sample}
	\parbox[1]{\textwidth}{\scriptsize \vspace*{2ex}\textit{Note:}	 In each of the panels, orange circles indicate choices in 2013 and
		blue circles indicate choices in 2016. The solid orange line represents CES
		estimation in 2013 and a dotted blue line represents the CES estimation in
		2016. These graphical comparisons provide some indication of goodness-of-fit
		for the four subjects we used for illustrative purposes. These figures are
		difficult to see in small black and white format. The scatterplots for the
		full set of subjects are available upon request.}
\end{figure}


\begin{figure}[p]
	\caption{The distributions of GARP violations}
	\begin{center}
	
		\includegraphics[width=1\textwidth]{tables-and-figures/figure2.pdf}
	
	\end{center}
	\label{fig:ratio}
	\parbox[1]{\textwidth}{\scriptsize \vspace*{2ex}\textit{Note:}	The fraction of subjects whom the minimum CCEI for the separate
		datasets (black) and the CCEI for the combined dataset (gray) are above
		different critical values. The CCEI provides a summary statistic of the
		overall consistency of the data with GARP. The closer the CCEI is to one,
		the smaller the perturbation of the budget constraints required to remove
		all violations and thus the closer the data are to satisfying GARP.}
\end{figure}





\begin{landscape}
	\centering
	\begin{figure}[p]
		\caption{The relationship between the distributional preferences in
			2013 and 2016 at the individual-level}
		\begin{small}
			\begin{center}
				\begin{adjustwidth}{-1cm}{-1cm}
					\includegraphics[scale=0.6]{tables-and-figures/summarize-alpha-changes.pdf}\includegraphics[scale=0.6]{tables-and-figures/summarize-rho-changes.pdf}\includegraphics[scale=0.6]{tables-and-figures/summarize-tokenshare-changes.pdf}
				\end{adjustwidth}
			\end{center}
			\parbox[1]{1.8\textwidth}{\scriptsize \vspace*{2ex}\textit{Note:}	Left panel: A scatterplots of the estimated fair-mindedness ($\hat{%
					\alpha}_{n}$) in 2013 and 2016. Middle panel: A scatterplots of the
				estimated equality-efficiency orientation ($\hat{\rho}_{n}$) in 2013 and
				2016. Right panel: the expenditure share of the tokens kept $p_{s}\pi _{s}$
				in 2013 and 2016. Blue triangles indicate statistically significant changes,
				while green circles indicate changes that are not statistically significant.
				The line in each panel is a polynomial regression plot with 95\% confidence intervals. The histograms display the distribution
				of parameters in 2013 and 2016. These figures are difficult to see in small
				black and white format.}
		\end{small}
		\label{fig:ccei}
	\end{figure}
\end{landscape}
\newpage
\clearpage


\begin{landscape}
	\begin{table}[h]
		\begin{adjustwidth}{-0.5in}{-0.5in}
			\caption{Comparing the ALP subjects with the U.S. population}
			\begin{footnotesize}
				\begin{center}
					%\scalebox{0.5}{
					\begin{tabular}{lccccccccccccccccc}
						\hline
						\hline
						& & & & & & &\\ [-2ex]
						& \textsc{Completed}       & \textsc{Completed }       & \textsc{Completed }  &          &  &  & \\ [0.6ex]
						& \textsc{Experiment}      & \textsc{Experiment }   & \textsc{Experiment}  & \textsc{Entire ALP}  & \textsc{Entire ALP}   & \textsc{US Adults} & \textsc{US Adults}\\ [0.6ex]
						& in 2013 & in 2016 (2013) & in 2016 (2016)  & 2013 &2016 &  2012 & 2016\\ [0.6ex]
						& (1) & (2) & (3) & (4) & (5)  & (6) & (7)\\ [0.6ex]
						\hline
						& & & & & & & &\\ [-2ex]
						\\
						\input{tables-and-figures/table1.tex}
						& & & & & & & & \\ [-2ex]
						\hline
						& & & & & & & \\ [1ex]
						\multicolumn{8}{p{22.0cm}}{\scriptsize{\textit{Note:} Column (1) includes the 1,002 ALP respondents who completed the
								experiment in 2013. In columns (2) and (3), include the data for the
								subsample of 687 respondents who also completed the experiment in 2016 based
								on the 2013 and 2016 ALP questionnaire, respectively. Columns (4) and (5)
								compare our experimental subjects to the entire ALP sample in 2013 and 2016
								and columns (6) and (7) compare them to the American Community Survey (ACS)
								conducted in 2012 and 2016. The ACS interviewed about 2.4 and 2.2 million
								respondents in 2012 and 2016, respectively. Averages are weighted to
								represent the adult population of the U.S. Place of residency is classified
								according to the Census Bureau regions: Northeast, Midwest, South and West.}}\\
					\end{tabular}
				\end{center}
				\label{tab:summstats}
			\end{footnotesize}
		\end{adjustwidth}
	\end{table}
\end{landscape}


\begin{table}[h]
	\begin{adjustwidth}{-0.5in}{-0.5in}
		\caption{Comparing the household income of the ALP subjects and the U.S. Population}
		\begin{footnotesize}
			\begin{center}
				%\scalebox{0.5}{
				\begin{tabular}{lcccc}
					\hline
					\hline
					& & & & \\ [-2ex]
					& \textsc{Completed }       & \textsc{Completed }  &          &   \\ [0.6ex]
					& \textsc{Experiment }   & \textsc{Experiment}   & \textsc{US Adults} & \textsc{US Adults}\\ [0.6ex]
					& in 2016 (2013) & in 2016 (2016)  &  2012 & 2016\\ [0.6ex]
					& (1) & (2) & (3) & (4)\\ [0.6ex]
					\hline
					& & & & \\ [-2ex]
					\input{tables-and-figures/table2.tex}
					& & & &  \\ [-2ex]
					\hline
					& & & &  \\ [-2ex]
					\multicolumn{5}{p{12.0cm}}{\scriptsize{\textit{Note:} Columns (1) and (2) report the  household income of the subsample of
							687 respondents who also completed the experiment in 2016 based on the 2013
							and 2016 ALP questionnaire, respectively. Columns (3) and (4) report the
							household income in the American Community Survey (ACS) conducted in 2012
							and 2016. }}\\
				\end{tabular}
			\end{center}
			\label{tab:summstats}
		\end{footnotesize}
	\end{adjustwidth}
\end{table}



\begin{table}[p]
	\caption{The relationship between the distributional preferences in
		2013 and 2016}
	%	\begin{small}
	\begin{center}
		%\begin{adjustwidth}{-1cm}{-1cm}
		\begin{tabular}{lccccccc}
			\hline \hline
			& & & & & \\ [-1.9ex]
			
			\input{tables-and-figures/table3.tex}
%			\hline
			& & & & & \\ [-1.9ex]
		\end{tabular}
		%\end{adjustwidth}
	\end{center}
	%	\end{small}
	\parbox[1]{1\textwidth}{\scriptsize \vspace*{2ex}\textit{Note:} Robust standard errors in parentheses; *, **, and *** indicate 10, 5,
		and 1 percent significance levels, respectively. All specifications
		estimated via OLS. The dependent variable in columns (1)-(2) is
		fair-mindedness ($\hat{\alpha}_{n}$) in 2016 and in columns (3)-(4) is
		equality-efficiency orientation ($\hat{\rho}_{n}$) in 2016. The independent
		variables are the parameter estimates, $\hat{\alpha}_{n}$ and $\hat{\rho}_{n}
		$, and the CCEI score in 2013. In columns (3) and (4), we report the results
		with $\hat{\rho}_{n}$ in 2016 as the dependent variable for the 517 (75.3\%)
		fair-minded subjects ($\hat{\alpha}_{n}<1$) in both 2013 and 2016. We obtain
		similar results if we also include the $\hat{\rho}_{n}$ estimates of
		self-interested subjects ($\hat{\alpha}_{n}=0$). These results are presented
		in Appendix Table A1.}\\
	\label{tab:ces-stability-regs}
\end{table}


\newpage
\clearpage

\begin{table}[htbpp]
	\caption{The relationship between distributional preferences, economic
		circumstances and political preferences}
	\begin{small}
		\begin{center}
			\begin{adjustwidth}{-3cm}{-3cm}
				\begin{tabular}{lccccccccc}
					\hline \hline
					& & & & & &  & &\\ [-1.2ex]
		
					
					\input{tables-and-figures/table4.tex}
					& & & & & & & & \\ [-1.9ex]
					\hline
					& & & & & & &  & \\ [-1.2ex]
					
				\end{tabular}\\
			
				\footnotesize{Robust standard errors in parentheses; *, **, and *** indicate 10, 5,
					and 1 percent significance levels, respectively. All specifications
					estimated via OLS. The dependent variable in columns (1)-(4) is
					fair-mindedness ($\hat{\alpha}_{n}$) in 2016 and in columns (5)-(8) is
					equality-efficiency orientation ($\hat{\rho}_{n}$) in 2016. The independent
					variable of interest in column (1)-(2) and (5)-(6) is a variable
					which takes on values of ${-1,0,1}$ based on whether the subject's household
					income quartile decreased, stayed the same, or increased, respectively. The
					independent variable of interest in column (3)-(4) and (7)-(8) is an
					indicator variable which takes on values of ${-1,0,1}$ based on whether the
					subject shifted to voting Republican, did not change party (or has missing
					data on voting), or shifted to voting Democrat, respectively. Q2, Q3 and Q4
					are indicator variables for the subject's household income quartile in 2013
					(Q1 is the omitted category). In columns (5)-(8), we report the results with
					$\hat{\rho}_{n}$ in 2016 as the dependent variable for the 517 (75.3\%)
					fair-minded subjects ($\hat{\alpha}_{n}<1$) in both 2013 and 2016. We obtain
					similar results if we also include the $\hat{\rho}_{n}$ estimates of
					self-interested subjects ($\hat{\alpha}_{n}=0$). These results are presented
					in Appendix Table A2. Individual demographic characteristics include
					gender, ethnicity, age and educational attainment. We provide the full
					regression output including individual demographic controls in Appendix
					Table A3. }
				
				
			\end{adjustwidth}
		\end{center}
		
	\end{small}
	\label{tab:ces-change-regs}
	%	\parbox[1]{1.1\textwidth}{\scriptsize \vspace*{1ex}\textit{Note:}
\end{table}

% ----------------------------------- %
%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% APPENDIX A: ADDITIONAL TABLES AND FIGURES %%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%
% ----------------------------------- %
\newpage
\clearpage
\section*{Appendix A}
%% reset Figure counter
\setcounter{figure}{0}
\setcounter{table}{0}
\renewcommand{\thefigure}{A\arabic{figure} } % set caption label style to A.1
\renewcommand{\thetable}{A\arabic{table} } % set caption label style to A.1

\begin{table}[htbp]
	\caption{The relationship between the distributional preferences in
		2013 and 2016, Full Sample}
	%	\begin{small}
	\begin{center}
		%\begin{adjustwidth}{-1cm}{-1cm}
		\begin{tabular}{lccccccc}
			\hline \hline
			& & & & & \\ [-1.9ex]
			
			\input{tables-and-figures/table_a1.tex}
			& & & & & \\ [-1.9ex]
		\end{tabular}
		%\end{adjustwidth}
	\end{center}
	%	\end{small}
	\parbox[1]{1\textwidth}{\scriptsize \vspace*{2ex}\textit{Note:} \footnotesize{The regression results reported in Table 4, also including the $\hat{\rho}_{n}$ estimates of self-interested subjects ($\hat{\alpha}_{n}=0$) in columns (5)-(8). Robust standard errors in parentheses; *, **, and *** indicate 10, 5, and 1 percent significance levels, respectively. All specifications estimated via OLS. The dependent variable in columns (1)-(2) is fair-mindedness ($\hat{\alpha}_{n}$) in 2016 and in columns (3)-(4) is equality-efficiency orientation ($\hat{\rho}_{n}$) in 2016. The independent variables are the parameter estimates, $\hat{\alpha}_{n}$ and $\hat{\rho}_{n}$, and the CCEI score in 2013.}}\\
	\label{tab:ces-stability-regs}
\end{table}

\begin{table}[htbp]
	\caption{The relationship between distributional preferences, economic
		circumstances and political preferences}
	\begin{small}
		\begin{center}
			%	\begin{adjustwidth}{-1.9cm}{-1.9cm}
			\scalebox{0.7}{
				\begin{tabular}{lccccccccc}
					\hline \hline
					& & & & & & & & & \\ [-1.9ex]

					\input{tables-and-figures/table_a2.tex}
					& & & & & \\ [-1.9ex]
				\end{tabular}
			}
			%	\end{adjustwidth}
			\parbox[1]{1.1\textwidth}{\scriptsize \vspace*{1ex}\textit{Note:}The regression results reported in Table 5, also including the $\hat{ \rho}_{n}$ estimates of self-interested subjects ($\hat{\alpha}_{n}=0$) in columns (5)-(8). Robust standard errors in parentheses; *, **, and *** indicate 10, 5, and 1 percent significance levels, respectively. All specifications estimated via OLS. The dependent variable in columns (1)-(4) is fair-mindedness ($\hat{\alpha}_{n}$) in 2016 and in columns (5)-(8) is equality-efficiency orientation ($\hat{\rho}_{n}$) in 2016. The independent variable of interest in column (1)-(2) and (5)-(6) is a variable which takes on values of ${-1,0,1}$ based on whether the subject's household income quartile decreased, stayed the same, or increased, respectively. The independent variable of interest in column (3)-(4) and (7)-(8) is a variable which takes on values of ${-1,0,1}$ based on whether the subject shifted to voting Republican, did not change party (or has missing data on voting), or shifted to voting Democrat, respectively. Q2, Q3 and Q4 are indicator variables for the subject's household income quartile in 2013 (Q1 is the omitted category).}\\
			
		\end{center}
		
	\end{small}
	\label{tab:ces-change-regs}
	
\end{table}


\begin{table}[htbpp]
	\caption{The relationship between distributional preferences, economic
		circumstances and political preferences, Full Sample}
	\begin{small}
		\begin{center}
			\begin{adjustwidth}{-3cm}{-3cm}
				\begin{tabular}{lccccccccc}
					\hline \hline
					& & & & & &  & &\\ [-1.2ex]
					%	\textit{Dependent variable:} &  \multicolumn{3}{c}{$\hat{\alpha}_n$ \textsc{in 2016}} & %& \multicolumn{3}{c}{ $\hat{\rho}_n$ \textsc{in 2016}} \\
					%\\ [0.6ex] \cline{2-3} \cline{6-8}
					%& & & & & & & & \\ [-1.2ex]
					
					\input{tables-and-figures/table_a3.tex}
					& & & & & & & & \\ [-1.9ex]
					\hline
					& & & & & & &  & \\ [-1.2ex]
					
				\end{tabular}
				%		\parbox[1]{1.1\textwidth}{\scriptsize \vspace*{1ex}\textit{Note:}
				\footnotesize{A full regression output including individual demographic controls of the results presented in Table 5. Robust standard errors in parentheses; *, **, and *** indicate 10, 5, and 1 percent significance levels, respectively. All specifications estimated via OLS. The dependent variable in columns (1)-(4) is fair-mindedness ($\hat{\alpha}_{n}$) in 2016 and in columns (5)-(8) is equality-efficiency orientation ($\hat{\rho}_{n}$) in 2016. The independent variable of interest in column (1)-(2) and (5)-(6) is a variable which takes on values of ${-1,0,1}$ based on whether the subject's household income quartile decreased, stayed the same, or increased, respectively. The independent variable of interest in column (3)-(4) and (7)-(8) is a variable which takes on values of ${-1,0,1}$ based on whether the subject shifted to voting Republican, did not change party (or has missing data on voting), or shifted to voting Democrat, respectively. Q2, Q3 and Q4 are indicator variables for the subject's household income quartile in 2013 (Q1 is the omitted category). In columns (5)-(8), we report the results with $\hat{\rho}_{n}$ in 2016 as the dependent variable for the 517 (75.3\%) fair-minded subjects ($\hat{\alpha}_{n}<1$) in both 2013 and 2016. We obtain similar results if we also include the $\hat{\rho}_{n}$ estimates of self-interested subjects ($\hat{\alpha}_{n}=0$).}
				
				
			\end{adjustwidth}
		\end{center}
		
	\end{small}
	\label{tab:ces-change-regs}

\end{table}


\newpage
\clearpage



\end{document}
